The objective of this study is an in-depth examination of the health and health services use of the 7,447 HRS/AHEAD respondents aged 70 years old or older at baseline. We will analyze the public use data from the baseline (1993) and the 1995, 1998, 2000, 2002, and 2004 follow-up interviews merged with data from three of the restricted access files: Medicare claims, National Death Index, and geocodes. With the geocode files we will link State and county level data using the Area Resource File, and at the local access level using GIS. Our four specific aims are to: (1) examine the risk of all-cause hospitalization using self reports, using Medicare claims, and their correspondence; (2) assess the risk for eight specific major morbid events and their sequelae; (3) evaluate the risks for preventable or ambulatory care sensitive hospital episodes; and (4) model the late life course trajectories of hospitalization patterns using a very promising typology developed by the investigative team. Our theoretical framework includes sociodemographic characteristics, socioeconomic and other access factors, health beliefs and lifestyle, health status, and health services use. Embedded in this framework are four explicit hypotheses. First, when modeling all-cause hospitalization, we expect that the market structure and practice pattern factors will be the most salient predictors, followed by the local access measures. Second, we expect that the epidemiologic risk factors for each of the 8 common, morbid conditions will be the most salient factors when modeling those conditions, followed by the market structure and practice pattern measures, and the continuity of interpersonal primary care. Third, when modeling preventable or ambulatory care sensitive hospitalizations, we expect the most salient predictors will be education, continuity of interpersonal primary care, and the supply of primary care health care services in the local access area. Finally, when modeling post-baseline late life course trajectories of hospital utilization patterns, we expect that pre-baseline patterns will be the most salient predictors, followed by market structure and practice patterns, local access factors, and functional and socioeconomic status. Estimation techniques will principally rely on multilevel event history models, such as binomial and multinomial logistic regression, and proportional hazards models. Linear and non-linear regression and random effects models will also be used where appropriate. [unreadable] [unreadable]